Flattened, rotating disks of cool dust and gas extending for tens to hundreds of astronomical units are found around almost all low-mass stars shortly after their birth. These disks generally persist for several million years, during which time some material accretes onto the star, some is lost through outflows and photoevaporation, and some condenses into centimeter- and larger-sized bodies or planetesimals. Through observations mainly at IR through millimeter wavelengths, we can determine how common disks are at different ages; measure basic properties including mass, size, structure, and composition; and follow their varied evolutionary pathways. In this way, we see the first steps toward exoplanet formation and learn about the origins of the Solar System. This review addresses observations of the outer parts, beyond 1 AU, of protoplanetary disks with a focus on recent IR and (sub)millimeter results and an eye to the promise of new facilities in the immediate future.

http://www.ifa.hawaii.edu/users/cieza/WC2011_ARAA.pdf

Protoplanetary Disks and Their Evolution

Hubble Space Telescope (HST) images of HK Tauri reveal that the companion star in this 24 (340 AU) pre-main-sequence binary system is an entirely nebulous object at visual wavelengths. HK Tau/c appears as two elongated reflection nebulosities separated by a dark lane. Near-infrared adaptive optics observations made at the Canada-France-Hawaii Telescope show a similar morphology and no directly visible star at λ≤2.2 μm. HK Tau/c is strikingly similar to scattered light models of an optically thick circumstellar disk seen close to edge-on and to the HST images of HH 30. HK Tau/c is therefore the first disk to be clearly resolved around an individual star in a young binary system. The disk properties have been constrained by fitting model reflection nebulae to the HST images. The disk has a radius of 105 AU, an inclination of about 5°, a scale height of 3.8 AU at r=50 AU, and is flared. The absence of a point source in the near-IR requires AV>50 mag toward the unseen central star. The thickness of the dark lane establishes a disk mass near 10-4 M☉ (~0.1 MJupiter) of dust and gas, if the dust grains have interstellar properties and remain fully mixed vertically. With the observed disk radius equal to one-third of the projected separation of the binary, there is a strong possibility that tidal truncation of the circumsecondary disk has occurred in this system.

An Edge-on Circumstellar Disk in the Young Binary System HK Tauri

Language models (LMs) exhibit remarkable abilities to solve new tasks from just a few examples or textual instructions, especially at scale. They also, paradoxically, struggle with basic functionality, such as arithmetic or factual lookup, where much simpler and smaller models excel. In this paper, we show that LMs can teach themselves to use external tools via simple APIs and achieve the best of both worlds. We introduce Toolformer, a model trained to decide which APIs to call, when to call them, what arguments to pass, and how to best incorporate the results into future token prediction. This is done in a self-supervised way, requiring nothing more than a handful of demonstrations for each API. We incorporate a range of tools, including a calculator, a Q\&A system, two different search engines, a translation system, and a calendar. Toolformer achieves substantially improved zero-shot performance across a variety of downstream tasks, often competitive with much larger models, without sacrificing its core language modeling abilities.

Toolformer: Language Models Can Teach Themselves to Use Tools

Language is essentially a complex, intricate system of human expressions governed by grammatical rules. It poses a significant challenge to develop capable AI algorithms for comprehending and grasping a language. As a major approach, language modeling has been widely studied for language understanding and generation in the past two decades, evolving from statistical language models to neural language models. Recently, pre-trained language models (PLMs) have been proposed by pre-training Transformer models over large-scale corpora, showing strong capabilities in solving various NLP tasks. Since researchers have found that model scaling can lead to performance improvement, they further study the scaling effect by increasing the model size to an even larger size. Interestingly, when the parameter scale exceeds a certain level, these enlarged language models not only achieve a significant performance improvement but also show some special abilities that are not present in small-scale language models. To discriminate the difference in parameter scale, the research community has coined the term large language models (LLM) for the PLMs of significant size. Recently, the research on LLMs has been largely advanced by both academia and industry, and a remarkable progress is the launch of ChatGPT, which has attracted widespread attention from society. The technical evolution of LLMs has been making an important impact on the entire AI community, which would revolutionize the way how we develop and use AI algorithms. In this survey, we review the recent advances of LLMs by introducing the background, key findings, and mainstream techniques. In particular, we focus on four major aspects of LLMs, namely pre-training, adaptation tuning, utilization, and capacity evaluation. Besides, we also summarize the available resources for developing LLMs and discuss the remaining issues for future directions.

A Survey of Large Language Models

We present results from a 2300 arcmin^2 survey of the Orion A molecular cloud at 450 and 850 micron using the Submillimetre Common-User Bolometer Array (SCUBA) on the James Clerk Maxwell Telescope. The region mapped lies directly south of the OMC1 cloud core and includes OMC4, OMC5, HH1/2, HH34, and L1641N. We identify 71 independent clumps in the 850 micron map and compute size, flux, and degree of central concentration in each. Comparison with isothermal, pressure-confined, self-gravitating Bonnor-Ebert spheres implies that the clumps have internal temperatures T_d ~ 22 +/- K and surface pressures log (k^-1 P cm^-3 K) = 6.0 +/- 0.2. The clump masses span the range 0.3 - 22 Msun assuming a dust temperature T_d ~ 20 K and a dust emissivity kappa_850 = 0.02 cm^2 g^-1. The distribution of clump masses is well characterized by a power-law N(M) propto M^-alpha with alpha = 2.0 +/- 0.5 for M > 3.0 Msun, indicating a clump mass function steeper than the stellar Initial Mass Function. Significant incompleteness makes determination of the slope at lower masses difficult. A comparison of the submillimeter emission map with an H_2 2.122 micron survey of the same region is performed. Several new Class 0 sources are revealed and a correlation is found between both the column density and degree of concentration of the submillimeter sources and the likelihood of coincident H_2 shock emission.

Large Area Mapping at 850 Microns. V. Analysis of the Clump Distribution in the Orion A South Molecular Cloud

Driven by the goal of eradicating language barriers on a global scale, machine translation has solidified itself as a key focus of artificial intelligence research today. However, such efforts have coalesced around a small subset of languages, leaving behind the vast majority of mostly low-resource languages. What does it take to break the 200 language barrier while ensuring safe, high quality results, all while keeping ethical considerations in mind? In No Language Left Behind, we took on this challenge by first contextualizing the need for low-resource language translation support through exploratory interviews with native speakers. Then, we created datasets and models aimed at narrowing the performance gap between low and high-resource languages. More specifically, we developed a conditional compute model based on Sparsely Gated Mixture of Experts that is trained on data obtained with novel and effective data mining techniques tailored for low-resource languages. We propose multiple architectural and training improvements to counteract overfitting while training on thousands of tasks. Critically, we evaluated the performance of over 40,000 different translation directions using a human-translated benchmark, Flores-200, and combined human evaluation with a novel toxicity benchmark covering all languages in Flores-200 to assess translation safety. Our model achieves an improvement of 44% BLEU relative to the previous state-of-the-art, laying important groundwork towards realizing a universal translation system. Finally, we open source all contributions described in this work, accessible at https://github.com/facebookresearch/fairseq/tree/nllb.

No Language Left Behind: Scaling Human-Centered Machine Translation

We present new images and proper-motion measurements of irradiated outflows from young stars in the outskirts of the Orion Nebula obtained with the Advanced Camera for Surveys onboard the Hubble Space Telescope. The images reveal many new outflows and new details in previously detected flows. The large-scale bipolar flows such as HH 502, 505, 874, and 876 contain multiple shocks whose velocities systematically decrease with increasing distance from their sources as the flows bend away from the core of the Orion Nebula. We consider several mechanisms for bending jets, including radiation pressure, the rocket effect on a neutral jet, and the impact of a side wind. While mild bends may be explained by either radiation pressure or the rocket effect, the extreme bends of some Orion jets and LL Ori–type bow shocks are best explained by the interaction of these outflows with a large-scale flow from the nebular core. The jet Hα emission measures and geometry are used to estimate their mass-loss rates, which range from about ~ 10-9 to 10-6 M⊙ yr-1. Many of the quasi-parabolic bow shocks that wrap around young stars in Orion (the LL Ori–type objects) exhibit large gaps between the observed jets and the parabolic fronts facing the core of the nebula. These may indicate the additional action of a wide-angle outflow component such as a T Tauri stellar wind, a proplyd photoablation flow, or possibly a photoablation flow from the jet itself. The Hα surface brightness and radii of the LL Ori fronts are used to estimate that the Vw products of the wide-angle flow components have values around 10-6 M⊙ yr-1 km s-1. The side-wind density and velocity are also constrained. Outside the nebular core, the greatest concentration of bent jets and LL Ori–type parabolic fronts is located south and west of the Trapezium. The nonuniform, clumpy spatial distribution of jets, outflow sources, and proplyds in the outskirts of the Orion Nebula indicate that star formation occurred in small hierarchical clusters.

Irradiated and Bent Jets in the Orion Nebula

We report the detection of several new circumstellar disks seen in silhouette against background nebular light in the outskirts of the Orion Nebula and the neighboring H II region M43. These were detected as part of our Hα survey of Orion with the Advanced Camera for Surveys on board the Hubble Space Telescope. Several of the disks show bipolar reflection nebulae, microjets, or pronounced temporal variability of their central stars. The relatively large fraction of bipolar reflection nebulae and microjets in our sample may be a selection effect caused by the faint nebular background far from the Trapezium. Two disks in our sample are large and particularly noteworthy: A nearly edge-on disk, d216-0939, is located several arcminutes northwest of M43 and resembles the famous HH 30 disk/jet system in Taurus. It drives the 0.15 pc long bipolar outflow HH 667 and exhibits a remarkable asymmetric reflection nebula caused by the tilt of the flared disk. With a diameter of ~26 (1200 AU), d216-0939 is as large as the giant edge-on silhouette disk d114-426 in the core of the Orion Nebula. The large disk d253-1536 is located in a binary system embedded within an externally ionized giant proplyd in M43. The disk exhibits distortions that we attribute to tidal interactions with the companion star. The bipolar jet HH 668 emerges from the proplyd ionization front in a direction orthogonal to the disk and can be traced to the young star embedded within it. A bow shock lies 54'' south of this binary system along the outflow axis. Proper motions over a 1.4 yr baseline confirm that these emission knots are indeed moving away from d253-1536 with speeds as high as ~330 km s-1 in the HH 668 microjet and slower motion farther from the star.

New Silhouette Disks with Reflection Nebulae and Outflows in the Orion Nebula and M43

We report the detection of several new circumstellar disks seen in silhouette in the outskirts of the Orion nebula and M43, detected as part of our Halpha survey of Orion with the HST/ACS. Several of the disks show bipolar reflection nebulae, microjets, or temporal variability. Two disks in our sample are large and particularly noteworthy: A nearly edge-on disk, d216-0939, is located several arcminutes northwest of M43 and resembles the famous HH30 disk/jet system in Taurus. It drives the 0.15 pc long bipolar outflow HH667, and exhibits a remarkable asymmetric reflection nebula. With a diameter of 1200 AU, it is as large as the giant edge-on silhouette disk d114-426 in the core of the Orion Nebula. The large disk d253-1536 is located in a binary system embedded within an externally-ionized giant proplyd in M43. The disk exhibits distortions which we attribute to tidal interactions with a companion. The bipolar jet HH668 emerges orthogonal to the disk, and a bow shock lies 54'' south of this binary system along the outflow axis. Proper motions over 1.4 yr confirm that these emission knots are moving away from d253-1536, with speeds as high as 330 km/s in the HH668 microjet, and slower motion farther from the star.

We present Hubble Space Telescope Advanced Camera for Surveys images of the giant proplyd 181-826 in the southern Orion Nebula. This object exhibits a variety of known proplyd properties—an optically visible central star, a silhouette disk, a bright proplyd ionization front, and a bipolar microjet emerging along the disk axis that powers a larger Herbig-Haro flow. The proplyd ionization front is remarkable because of its large size (a radius of about 800 AU) and because of its corrugated structure consisting of a set of coaxial rings with the same orientation as the embedded silhouette disk. The rings are centered on the disk/jet axis, and we discuss possible mechanisms for their formation. The silhouette disk has an outer radius of about 160 AU, and its axis is inclined by an angle of about 60° with respect to our line of sight. A prominent reflection nebula protrudes along the disk axis toward the south, marking the base of the approaching outflow, and a faint knot on the north side of the disk indicates that the nebula is bipolar. Fabry-Perot images reveal a high-velocity blueshifted jet with a mass-loss rate of about 10-8 M⊙ yr-1 emerging along the disk axis. It has a velocity (corrected for the outflow inclination angle) of about 160 km s-1 and can be traced for about 1' toward the south. A chain of three lower velocity blueshifted bow shocks, collectively known as HH 540, extend several arcminutes south of 181-826 along the same axis, tracing a large-scale outflow powered by this jet.

Daniel Licht

Papers

No Language Left Behind: Scaling Human-Centered Machine Translation

Irradiated and Bent Jets in the Orion Nebula

New Silhouette Disks with Reflection Nebulae and Outflows in the Orion Nebula and M43

New Silhouette Disks with Reflection Nebulae and Outflows in the Orion Nebula and M43

Outflows in the Orion Nebula: HH 540 from the Beehive Proplyd*